This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Martínez, J. L. Articles by Baquero, F. Search for Related Content PubMed PubMed Citation Articles by Martínez, J. L. Articles by Baquero, F.

Interactions among Strategies Associated with Bacterial Infection: Pathogenicity, Epidemicity, and Antibiotic Resistance

Departamento de Biotecnología Microbiana, Centro Nacional de Biotecnología,¹ Servicio de Microbiología, Hospital Ramón y Cajal, Madrid, Spain²

SUMMARY

INTRODUCTION

GENETIC LINKAGE OF RESISTANCE AND VIRULENCE

Common Mechanisms Involved in Virulence and Resistance

Costs of and Compensations for Virulence and Resistance

Evolution and Dissemination of Genes Involved in Virulence and Resistance

Mutation and recombination.

Plasmids.

Transposons.

Phages.

Gene cassettes.

Phenotypic Adaptation in Virulence and Antibiotic Resistance

Environmental signaling in virulence and antibiotic resistance.

Antibiotics as effectors of bacterial virulence.

Cross talk between virulence and antibiotic resistance regulons.

THE CROSSROADS OF VIRULENCE, EPIDEMICITY, AND ANTIBIOTIC RESISTANCE

Are Antibiotic-Resistant Bacteria More or Less Virulent?

Are Antibiotic-Resistant Bacteria More or Less Epidemic?

Are Virulent Bacteria More Resistant to Antibiotics?

Are Epidemic Bacteria More Resistant to Antibiotics?

RESISTANCE AND VIRULENCE: AN INTEGRATED VIEW

Resistance as a Colonization Factor in the Treated Host

Opportunistic Infections

Does Virulence Affect Exposure to Antibiotics?

In host spatial location and exposure to antibiotics.

Normal microbiota versus pathogenic bacteria.

POPULATION STRUCTURE

Bacterial Variation Facing Stressful Environments

Stress-induced transient increases in mutation rates.

Hypermutable (mutator) strains.

Clonal Structure

Resistance and virulence decrease bacterial diversity.

Taxonomic Implications of Changes in Resistance and Virulence

Misclassification as new species.

Emergence of novel microbial species.

FUTURE EVOLUTION IN AN ECOLOGICAL PERSPECTIVE

Major Factors

Sanitation and hygienic measures (mainly drinking water and sewage management).

Preservation and eventual bioremediation of normal ecosystems in human, animal, and environmental microbiota.

Number of contacts between microbes from infected and noninfected hosts (human and animal) and between infected people and the environment.

Global environmental changes.

Human sociological movements.

Increase in susceptible hosts.

New technologies.

Intensive farming.

Viral and virus-like diseases.

New antimicrobial strategies.

Vaccination.

Bioterrorism.

Strategies against Virulence May Reduce Antibiotic Resistance

Reducing host-to-host transmission: hygiene and vaccination.

Antivirulence-directed therapeutic approaches.

Strategies Directed against Resistance May Reduce Bacterial Virulence

Drug diversification.

INFLUENCE ON EVOLUTION OF THE HOST

Changes in Normal Host Microbiota

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
The mechanisms involved in the virulence (defined as the relative capacity of a microbe to cause damage in a host [72, 73]) of pathogenic bacteria as well as those determining antibiotic resistance are important and widely studied topics in clinical microbiology. However, they have rarely been analyzed and integrated as we intend to do in the present review.

In terms of evolution and ecology, antibiotic resistance and virulence determinants share some basic characteristics. Since these determinants have been acquired by horizontal gene transfer from other organisms, many are examples of what has been termed "evolution in quantum leaps" (153). Also, most determinants serve to escape the action of antibacterial defense systems that have developed either by natural (host anti-infectious mechanisms) or cultural (antibiotics) evolution (123) to prevent infections. Both the natural anti-infective defenses and antibiotic treatments lead to stringent conditions for bacterial growth. In biology, any limiting condition for the majority is a golden opportunity for the minority. Those bacteria that are capable of surviving and multiplying under these conditions will gain access to organic spaces in which competition with other microorganisms is avoided (exclusive environments).

We might then assume that both virulence and antibiotic resistance are formally similar adaptive mechanisms selected to survive under stress conditions (either host invasion or antibiotic treatment). From an ecological point of view, both infective conditions and antibiotic treatments are evolutionary bottlenecks that tend to reduce microbial biodiversity, so that only a very specific subset of bacteria are capable of colonizing the host under those conditions (Fig. 1 ). There are several bacterial species that are able to grow at 37°C and are tolerant to the oxygen tension present in different parts of the human body. The fact that environmental microorganisms that are unable to produce disease in the healthy host frequently infect immunocompromised patients indicates that many organisms are ecologically compatible with the physicochemical conditions within the human body. The human body and its physicochemical conditions are then an ecological space that can be colonized by several microorganisms (182), frequently with an environmental origin (307). In the normal host, this potential for colonization is limited by the immune system, which actively impedes colonization of the human body by opportunistic pathogens. In the immunodepressed host, only antibiotic treatment maintains a small colonizable space in the human body (see below).

View larger version (44K): [in this window] [in a new window]   FIG. 1. Infection and antibiotic treatment are both stringent growing conditions. Several bacteria are able to grow at the temperature and oxygen tension of and using the nutrients present in the human body. However, only some are able to produce infection; this is shownin panel A. Several bacterial species (ovals are pathogens, circles are environmental ones) coexist outside the host. Some species are able to displace the commensal flora, traverse through different epithelia, resist the action of macrophages, antibodies, defensins (with squares), and all the anti-infectious mechanisms of the human body, to finally reach the target cells where the disease is produced. At any of these sequential bottlenecks, only some bacteria are selected, and their population is further amplified, so that, from the high variability of bacteria that could potentially produce the disease, only some are really pathogenic. In panel B, the same situation is analyzed but under antibiotic treatment (spheres). In this case, only a small proportion of the cells belonging to the infectious species (the antibiotic-resistant ones; blue ovals with a red line) are able to produce infection, so that the antibiotic-treated host is a more stringent ecosystem for the growth of bacteria. However, in the case of a debilitated person (panel C), the situation is somewhat different. In these patients, the indigenous microflora might be removed because of antibiotic use, the epithelial integrity may be impaired (intubated patients), the immune system can be abolished (immunocompromised patients), and even the target cells can change. Under these circumstances, some environmental species can produce infection (opportunistic pathogens), because growth conditions are less stringent. However, at least in the developed world, those patients are usually under heavy antibiotic treatment, frequently with a combination of antibiotics (golden and blue spheres), so that only those species with an intrinsically antibiotic-resistant phenotype can produce the infection. Under these circumstances, the main selective force is antibiotics, so that antibiotic treatment becomes a risk factor for some pathogens.

If new antibiotics and protocols are required for fighting infectious diseases, we must first understand the relationship between virulence, transmissibility, and antibiotic resistance. Thus, two essential topics are first reviewed in this work: the effect of acquisition of antibiotic resistance (and antibiotic treatment) on bacterial virulence, and analysis of whether pathogenicity and antibiotic resistance might prevail as linked phenotypes in the world of the future, if pathogens become antibiotic resistant as a consequence of intense antibiotic selective pressure.

GENETIC LINKAGE OF RESISTANCE AND VIRULENCE

Top Previous Next References

 
Virulent bacteria have acquired their phenotype through a long evolutionary course in close contact with their natural hosts. Most virulence determinants are either located in chromosomal gene clusters (pathogenicity islands) or harbored in genetic accessory elements such as plasmids and phages. This suggests that evolution from an avirulent way of life to pathogenicity frequently implies the acquisition of foreign pieces of DNA (125, 153, 157). Nevertheless, for the organism to be a real pathogen, these pathogenicity factors should be inserted in an organism ecologically compatible with the potential host. Moreover, in some cases it is not an acquisition but a deletion (virulence-associated "black holes") which is needed to become a pathogen (254, 271, 359).

Indeed, any change in lifestyle has a biological cost, as functions needed in one habitat may cause a burden in another habitat and could therefore be counterselected. Acquisition of a virulence phenotype might then require the acquisition of some different pathogenicity islands and the loss of some chromosomal DNA regions. Therefore, the construction of a pathogen by the acquisition of specific pathogenic elements in ecologically compatible host-adapted bacterial genomes has probably occurred over a long evolutionary time.

In this regard, the record of infectious diseases (mainly epidemic ones) in human history and bioarchaeological analysis of the paleontological record indicate that we have been in contact with pathogens similar to those currently producing infections for a long time. In this way, it has been demonstrated that infectious diseases such as syphilis, tuberculosis, and other infections due to bacteria, viruses, and parasites were common in prehistoric men (420), particularly in the Neolithic age, when agriculture and farming ensured close contact between humans and between humans and animals. In some cases, however, such evolution has occurred in historical times, as the evolution from Yersinia pseudotuberculosis to Y. pestis, which was shortly followed by the first pandemic plagues (2). Obviously, the evolutionary outcome of pathogenic organisms tends to be limited in modern times by the anti-infectious repertoire developed by humans that includes hygienic measures, epidemiological controls, vaccines, and, significantly, antibiotics.

Conversely, acquisition and further spread of antibiotic resistance genes among pathogenic bacteria is a phenomenon that has occurred just in the last 50 years as a consequence of extensive antibiotic use for human therapy and animal farming. At first glance, pathogenicity and resistance should be unlinked phenomena. However, several examples indicate that this is not the situation for several bacterial pathogens. Antibiotic resistance and virulence genes can be linked (and then coselected) in the same replicon, or eventually a single determinant can be involved in both virulence and resistance. Some examples of these situations will be reviewed.

Let us first examine the effect of the spatial localization of bacteria in the human body. Several virulent bacteria base their pathogenic characteristics in an intracellular way of life during infection (125). Internalization may be required for the induction of inflammatory cytokines (166) and produce tissue damage through the induction of either necrotic (130) or apoptotic (167, 410) responses. A remarkable example is the apoptosis of macrophages induced by pathogens such as Shigella, which both avoid the antibacterial effect of those cells and trigger inflammation (427, 428). In some well-characterized examples, bacteria are able to travel from cell to cell without any significant contact with the extracellular milieu (43, 273), thus avoiding the immune system as well as contact with antibiotics that do not enter mammalian cells (12, 172). This provides the first clue as to how well-characterized virulence phenotypes can be related to antibiotic resistance.

Mammalian cells are poorly permeable to or easily exclude several families of antibiotics (12, 166). Even if the antibiotic enters the mammalian cells, intracellular growth might induce a transient antibiotic-resistant phenotype, a situation that has been described for Legionella pneumophila (35). The location within the host may alter in an unspecific way the susceptibility of the bacterial organism to antibiotics. Haemophilus influenzae grown in animals undergoes modifications in penicillin-binding proteins, as peptidoglycan metabolism is directly affected by the environment (93), and Salmonella peptidoglycan is dramatically altered for bacteria growing intracellularly (308). Although the effect of these changes on antibiotic susceptibility has not been analyzed extensively, it is possible that those metabolic changes may alter the activity of antibiotics such as beta-lactams against bacteria growing during infection. It is then possible that intracellular localization may allow bacteria to maintain a phenotype of antibiotic resistance.

In this respect, treatment with some antibiotics might select for intracellular clones. This could be the case for species in which some clones are able to invade mammalian cells and others are noninvasive; one such species is Pseudomonas aeruginosa (126). On the contrary, some antibiotics, such as macrolides and fluoroquinolones, are actively accumulated by human cells (262). In this case, intracellular bacterial localization might eventually increase the antibiotic effect against bacterial pathogens, leading to better eradication and favoring the less invasive strains.

A similar situation of phenotypic antibiotic resistance can be observed for bacteria growing in biofilms. Bacterial biofilms are frequent in persistent infections (85), such as those associated with cystic fibrosis (354), chronic bronchitis, osteomyelitis (152), and foreign-body-associated infections. Bacteria growing in biofilms are more resistant than those under a planktonic way of life to the action of phagocytic cells' antibacterial activity (261) as well as the action of antibiotics (16, 83, 149, 370, 404). As in the previous case, antibiotics might select biofilm-forming organisms, thereby increasing the prevalence of chronic infections. Other mechanisms of virulence could prevent the action of antibiotics against bacteria. For instance, toxins leading to local necrosis or abscess formation certainly will decrease the local availability of antibiotics due to reduction in antibiotic arrival to the foci or because of local inactivation of the drugs by altered pH, free proteins, or DNA. In these cases, the mechanisms of pathogenicity may finally serve as mechanisms for antibiotic resistance.

A final example of a virulence mechanism with a role in antibiotic resistance is presented by Bordetella pertussis, the etiological agent of whooping cough. The cell wall of the virulent strains of this microorganism is infrequently susceptible to autolysis triggered by beta-lactam antibiotics; only avirulent B. pertussis strains are known to be lysed. It was demonstrated that this phenotypic tolerance and the antibiotic-induced autolytic activity are controlled by the vir locus, which determines phase transition in B. pertussis, a key element in the virulence properties of this bacterial species (389).

In previous examples, we have seen that a "virulent way of life" can contribute to an antibiotic resistance phenotype. Could antibiotic resistance determinants also contribute to bacterial virulence? Some examples suggest that this could be the case. Recently, one area of intense study of antibiotic resistance is the analysis of multidrug resistance (MDR) efflux pumps (280, 293). These determinants are able to extrude an ample range of substances that include antibiotics, solvents, dyes, and quorum-sensing signals (295, 423). Extrusion by the same pump of compounds as diverse as erythromycin, quinolones, beta-lactams, and ethidium bromide is a rule more than an exception.

Might these MDR determinants extrude compounds involved in the host defense be contributing to bacterial virulence? The answer is yes. For instance, adaptation to growth in the presence of bile salts is a prerequisite for any pathogen to colonize the intestinal tract. It has been reported that Escherichia coli extrudes bile salts (382) through the acrAB system, which was first characterized as an MDR determinant (6, 230); the same occurs in Salmonella enterica serotype Typhimurium. In fact, it has been suggested that extrusion of bile salts present in the intestinal ecosystem to which these bacteria are adapted is the function for which those MDR determinants were selected (230). Analysis of S. enterica serotype Typhimurium mutants with high susceptibility to bile salts has demonstrated that this higher susceptibility was a consequence of inactivation of the MDR determinant acrB. This inactivation leads to a reduced ability to colonize the intestinal tract in a murine infection model (199), which indicates that acrAB is involved in both antibiotic resistance and virulence.

Some MDR systems, such as Mtr from Neisseria gonorrhoeae (158) and SapA (291) from S. enterica serotype Typhimurium, can actively extrude both defensins and antibiotics, thus contributing to reducing antibiotic susceptibility as well as to increasing the number of virulence determinants of the organisms. In fact, it has been described that resistance to defensins is required for a full virulence phenotype in the case of Salmonella (154). The mechanisms of action of defensins resemble at least in part that of peptidic antibiotics currently used in clinical practice (180, 391). Thus, it is conceivable that they might share the same mechanisms for resistance. This situation has been described for polymixin B. The sensitivities to polycationic peptides were compared between two groups of Yersinia enterocolitica isolates, one with an environmental origin and the other from infections (38). Data indicated that pathogenic strains of Y. enterocolitica were resistant to defensins and also to polymixin B, showing a clear cross-linkage between antibiotic resistance and virulence as a consequence of the expression of a resistance determinant.

Not only MDR pumps but also biocide efflux determinants may play a role in bacterial virulence. It has been reported that the QacA pump of Staphylococcus aureus, involved in resistance to several organic cations (biocides included), has a role in resistance to thrombin-induced platelet microbicidal protein (197). Thus, the presence of the pump contributes to the survival of strains carrying this determinant at sites of endothelial damage as well as in experimental endovascular infections. We want to stress here that the mechanisms of extrusion of antibiotics are similar to those translocating some proteins involved in pathogenicity. The hemolysin export apparatus in Escherichia coli comprises the outer membrane channel trimmer protein TolC that is also involved in the AcrA-B multidrug efflux transport complex (279). The periplasmic proteins of both systems, AcrA and HlyD, have a similar structure and are able to interact with TolC (278).

In all previously discussed examples, a mechanism selected for bacterial virulence can also produce an antibiotic resistance phenotype. Selection for the most virulent strains might select for antibiotic resistance, and conversely, selection for antibiotic resistance might select the most virulent microorganisms. However, the opposite situation can also be found: antibiotic resistance may decrease bacterial virulence (see below). For instance, the KatG catalase-peroxidase activity is important for the survival of Mycobacterium tuberculosis in the host (216). Mutations that eliminate this activity prevent the activation of isoniazid and are the major cause of resistance to this drug in Mycobacterium spp. (316, 425). It might then be predicted that isoniazid-resistant Mycobacterium mutants might be less virulent than wild-type isoniazid-susceptible strains, at least in the case of M. tuberculosis (264) and Mycobacterium bovis (414). However, when isoniazid-resistant isolates were analyzed, such reduced virulence was not found. It turned out that M. tuberculosis pathogenic isolates accumulate compensatory mutations in the gene encoding the alkyl peroxidase AphC, which can substitute for KatG for survival inside the host (350).

We might speculate that acquisition of these determinants implied a cost in bacterial fitness when they were included in the genome of the previously nonvirulent bacteria. However, acquisition of resistance genes by pathogenic bacteria has occurred over the last 50 years, but acquisition and further evolution of structures as pathogenicity islands occurred thousands of years ago. Thus, the evolutionary time needed to acquire and optimize compensatory mutations to alleviate the effect on fitness of the acquisition of pathogenicity determinants has been much longer. Also, for plasmids carrying vir determinants, the costs of the acquisition of such determinants might be compensated for by mutations in other loci. Of note, it has been shown that the vir plasmid can be easily lost in vitro in Shigella (342). However, growth conditions in vivo should preserve the plasmid's presence during infection, otherwise Shigella would not be virulent anymore.

Acquisition of vir plasmids ensures the ability to colonize a different biological compartment, thus evading competition with other bacterial species and eventually reducing the potential fitness cost imposed by the new genetic element. The same has probably occurred for enteroinvasive E. coli isolates (335, 336). Recent work suggests that the invasive ability of those isolates has evolved in different chromosomal backgrounds, probably through the spread of plasmid-borne invasion genes, and the maintenance of invasive phenotypes in separate lineages suggests that this ability confers a selective advantage to invasive strains (247). Although the Ewald hypothesis (119, 120) is still controversial, it can be speculated that sanitation procedures, vaccination, and widespread antibiotic use, impeding host-to-host spread of vir plasmid-based pathogens, may reduce the overall pathogenic power of microorganisms.

Due to the rapid acquisition and fixation of compensatory mutations by bacteria, experiments to evaluate the biological cost of the expression of novel antibiotic resistance and virulence determinants are difficult to perform. Antibiotic resistance and virulence can be acquired either by horizontal transfer of antibiotic resistance genes (97) or by mutation (241). The presence of new plasmids and transposons in the bacterial genome has a relevant cost for the recipient bacteria (13). However, it has been demonstrated that under these circumstances, the cost is compensated for in few generations as the consequence of genetic change by the host, getting the specific plasmid-bacterial strain association more fit than the previous non-plasmid-containing bacterium (61).

Also, in the case of mutations leading to antibiotic resistance, some examples demonstrate the possibility of fitness reduction, which may also reduce bacterial virulence (see last example in previous section). For instance, current evidence supports the idea that highly fluoroquinolone-resistant Salmonella strains could be counterselected in the field (145). A substantial reduction of the virulent characteristics has been also described for antibiotic-resistant S. enterica serotype Typhimurium isolates (48). Although this may occur for some bacteria, it does not mean that it will always occur. In fact, an outbreak of quinolone-resistant S. enterica serotype Typhimurium DT104 has recently been described (268), and an increase in quinolone-resistant Campylobacter jejuni infections has been reported in Minnesota (356).

The emergence of compensatory mutations rapidly alleviates the biological cost of antibiotic resistance in S. enterica serotype Typhimurium isolates (45), and this could be the situation for the aforementioned outbreaks of quinolone-resistant bacteria. Noteworthy, the compensatory mutations that are selected in vivo and in vitro are different, reflecting the different growing conditions for bacteria during infection compared with microorganisms growing in vitro (49). This indicates that the metabolic requirements needed for infection are different from those for surviving in the environment. Therefore, acquisition of the characteristics required for infection and antibiotic resistance might make them less proficient for surviving in the environment and to be transmitted among different hosts.

It seems then that acquisition of novel virulence and antibiotic resistance traits has a cost in bacterial fitness, but the cost is rapidly compensated for due to the high plasticity of bacterial genomes and the huge populations of bacteria from which compensatory mutants can be selected (59, 209, 213). However, this situation may not always occur. Possibly, only those mechanisms that can be compensated for by mutations in other loci are selected because these are the only ones that can avoid fitness reduction. As stated by other authors (13), the effect of antibiotic resistance (and in a higher grade of virulence determinants) on bacterial fitness has been properly addressed in only a few studies. This topic needs to be investigated so that the biological potential of novel virulent and antibiotic-resistant bacteria may be predicted.

It is widely accepted by the plasmids and transposons that carry antibiotic resistance genes may have originated in antibiotic-producing organisms in order to avoid the deleterious effect of the antibiotic on themselves (39, 98, 239, 290, 407). Eventually, they could have further evolved in organisms in an ecological consortium with antibiotic producers. More recent works indicate that this hypothesis about the origin of resistance is only half true (11, 98). For instance, it is difficult to believe that chromosomal beta-lactamases (71, 217) as well as some aminoglycoside-inactivating enzymes (4, 201, 233) and a plethora of MDR determinants (280, 293) which are present in all isolates of a given bacterial species originated in the antibiotic producers. In fact, all these determinants must have a functional role other than antibiotic resistance, and this phenotype will only be a consequence of their primary physiological function (see reference 11 for a review of this concept).

In the case of virulence determinants, the current paradigm indicates that pathogenicity islands acquired through horizontal gene transfer are frequently responsible for the pathogenic properties of virulent bacterial species (125, 153, 157). Different from antibiotic resistance genes, the pathogenicity islands are more difficult to explain, because pathogenic bacterial ancestors carrying such gene clusters have not been detected. It has been proposed that pathogenicity islands might be relevant for the biodegradative properties of microorganisms (including decomposition of dead bodies), to kill living cells (to obtain food and reduce competition), and to live inside eukaryotic cells (such as amoeba, protozoa, plants, and animals) in natural environments (161).

Antibiotic resistance genes and virulence determinants might then play a different role in the original organisms from which they were transferred to pathogenic bacteria. However, once those determinants have been selected in the heterologous host, they confer a selective advantage and are fixed in the pathogenic bacterial populations, where they can evolve further and eventually be transferred to other bacterial species.

Mutation and recombination. Two different processes account for evolution of antibiotic resistance-virulence phenotypes: gene mutation and gene recombination (horizontal gene transfer included). The role of mutation in antibiotic resistance is well known (241). In the case of virulence, fewer examples of such a role have been described. However, mutational activation-inactivation of intrinsic genes might also contribute to a virulent phenotype in the case of opportunistic pathogens. As an example of cryptic virulence determinants activatable by mutations, laboratory isolates of E. coli are hemolytic (105) because of mutations in the hns (147) and fnr (315) genes. Whether this is just a laboratory curiosity and might actually have a role in the course of nosocomial infections by E. coli is a matter of speculation.

Some virulence determinants of P. aeruginosa, such as alginate production (60) and cytotoxicity, are downregulated. However, mutants in which the expression of such determinants is derepressed are frequently found during infection (240). This indicates that mutation probably plays a major role in the emergence of the different virulent phenotypes shown by P. aeruginosa clinical isolates. Pathogenic P. aeruginosa isolates can be broadly classified into two groups, those with a cytoinvasive phenotype and those with a cytotoxic phenotype. However, both kinds of isolates contain the genes required for invading epithelial cells. Invasive and cytotoxic strains differ in the expression of the genes under the control of an activator called ExsA (116). Mutations in the gene exsA, which encodes the activator, lead to not a cytotoxic but only an invasive phenotype (116, 126). Although both types of bacteria are virulent, they occupy different environments (inside and outside epithelial cells), so that one or the other phenotype might be selected in vivo by, for the moment, unknown processes.

Recombination also has an important role in the evolution of antibiotic resistance determinants. Progressive clustering of genes, leading to an operon structure, may have occurred to optimize and regulate the expression of ancient genes producing a resistance (frequently low-level resistance) phenotype (206). In a similar way, the construction of a pathogenicity island requires the recruitment of different genes that recombine to produce a single genetic element containing several genes with a common role, virulence. In other cases, intragenic recombination is essential to produce a resistant phenotype. The most conspicuous example of recombination in chromosomal DNA leading to antibiotic resistance is beta-lactam resistance produced by recombination of penicillin-binding-protein genes in Neisseria gonorrhoeae and Streptococcus pneumoniae (79, 235).

Similar recombinatorial processes may have influenced the evolution of virulence. However, unlike for antibiotic resistance, few examples of recent evolution of bacteria to a virulent phenotype have been described. One might be recombination between different sets of genes, which causes the rearrangements that can be observed in the capsular antigens of S. pneumoniae (78-80, 276) and can be considered a mechanism of defense against host immunity. Another could be the emergence and further dissemination of enterohemorrhagic E. coli O157:H7 strains (131). This E. coli serogroup has emerged as a relevant pathogen in the last 20 years. Enterohemorrhagic E. coli isolates contain virulence plasmids and pathogenicity islands similar to those found in Shigella spp. (70, 237). Although it was a matter of speculation whether these genetic determinants were acquired recently by E. coli, the recent sequencing of the genome of an E. coli O157:H7 strain (297) demonstrated that it contains as many as 1,387 new genes in comparison with the previously sequenced nonpathogenic laboratory strain E. coli K-12 (52).

We must be aware that acquisition of pathogenicity determinants by previously nonpathogenic organisms might occur in a similar way as it happens with antibiotic resistance determinants. For instance, the acquisition of virulence plasmids by bacteria forming part of the human indigenous flora and thus already well adapted for surviving inside their host (such as E. coli) might lead to the emergence of novel pathogens. In a similar way, the acquisition of virulence determinants by environmental microorganisms should produce novel pathogenic bacteria. Although the evolution of novel pathogens is possible, the selective pressure in favor of the selection of virulence determinants is not as strong as in the case of antibiotic resistance determinants. Therefore, an explosion of novel pathogens, such as has occurred with antibiotic-resistant bacteria, will most likely not happen in the near future, not only because the selective pressure for a pathogenic phenotype is less strong than for an antibiotic resistance phenotype, but because selection for pathogenicity in today's world might decrease with sanitation (119, 120) if the Ewald hypothesis is true.

A critical point to discuss here is that virulence and/or antibiotic treatment might increase the rate of bacterial variation. In the case of infection, this situation might contribute to increased mutation rates (310, 326) and even to the selection of hypermutator strains (see below), thus enhancing antibiotic resistance mutation (67, 208, 287, 380). In a similar way, antibiotic challenge might also produce hypermutable phenotypes (9, 318, 320), increasing the possibility of mutants overexpressing virulence determinants. Mechanisms favoring hypermutation may also facilitate recombination (253). Moreover, in vivo transfer of plasmids carrying antibiotic resistance genes and/or virulence determinants and recombination are probably favored during infection due to host signals that enhance gene transfer (260).

In this respect, the effect of antibiotics in inducing the transfer of plasmid and transposons has been demonstrated in vitro (387). It has been described that the same genes required to initiate infection of human macrophages by Legionella pneumophila are involved in plasmid mobilization (344). Recent results in our laboratory also suggest that bacterial expression of factors involved in cell-to-cell DNA transfer (likely antibiotic resistance plasmids) in some organisms, such as S. pneumoniae, may be triggered by inflammatory products (M. R. Baquero, unpublished results). It can then be suggested that bacteria are expected to evolve more rapidly inside the host and under antibiotic selective pressure, so that an infected patient under antibiotic therapy may act as an evolutionary accelerator.

Plasmids. It is well known that plasmids are major vectors for the dissemination of both antibiotic resistance and virulence determinants among bacterial populations. It is also clear that the presence of virulence and antibiotic resistance determinants in the same genetic element will produce coselection of both types of determinants. This applies as well for genes present in transposons, phages, and cassettes, which are discussed below. For bacteria carrying those linked determinants, the selection of an infective population will select for antibiotic resistance, and antibiotic selective pressure will select the virulence trait. Table 1 shows some published examples regarding this type of gene linkage. Among them, antibiotic resistance plasmids carrying, alone and in combination, genes encoding the synthesis of bacteriocins (272), siderophores (82, 168, 401), cytotoxins (34, 132), and adhesion factors (202) have been described.

The plasmid transfer proteins are sometimes chromosomally encoded, but they can have a role in both plasmid transfer and virulence. This could be the case for Legionella pneumophila. It has been described that the dot virulence genes encode a large putative membrane complex that functions as a secretion system that is able to transfer plasmid DNA from one cell to another (402). Mutations in the dot genes reduced both virulence and plasmid transfer. In the case of gram-positive bacteria, aggregation substances encoded by pheromone plasmids are involved in the clumping required for an efficient transfer of DNA by conjugation. It has also been described that Enterococcus faecalis aggregation substance promotes resistance to killing by human neutrophils in spite of phagocytosis and neutrophil activation (314) and enhances pathogenicity in a rabbit model (340) and thus, as said by R. Wirth, is "more than a plasmid collection mechanism" (416).

The analysis of plasmids from the preantibiotic era demonstrated that there have not been major changes in the families of plasmids present in pathogenic bacterial populations, but just recruitment of antibiotic resistance genes by formerly "antibiotic-susceptible" plasmids (95, 171, 179). Since plasmids encoding adaptive traits (eventually involved in the pathogenic lifestyle) certainly preceded antibiotic resistance, it could be suspected that many current resistance plasmids contain genes with a role in bacterial colonization and virulence. In fact, this has probably been the origin of plasmids containing virulence determinants and antibiotic resistance genes such as those listed in Table 1. We do not know, however, if this association is the rule for antibiotic resistance plasmids, because only a very small fraction of these determinants have been entirely sequenced.

Transposons. Transposons are also relevant for the dissemination of antibiotic resistance genes, either by integration in transferable plasmids or by direct conjugation and further integration in the bacterial chromosome. Also, transposons containing virulence determinants have been described. One example is the aerobactin operon (46, 142, 244). Aerobactin is a siderophore produced by several bacterial species (74). It has been proposed that aerobactin is a virulence factor with relevance for iron acquisition at the site of infection (103). Aerobactin genes have been found in the chromosomes of different bacterial species and in several different plasmids, a situation which indicates easy mobilization. The aerobactin operon forms part of a transposable element (102), and it has been proposed that it is part of a pathogenicity island in Shigella flexneri (403). Other virulence determinants found in transposable elements are the E. coli enterotoxin STII gene (170), the Shiga toxin operon, which has been found in a putative composite transposon in Shigella dysenteriae 1 (257), and the toxic shock toxin, carried by a family of mobile pathogenicity islands in Staphylococcus aureus (218).

Little is known about the association between antibiotic resistance and virulence determinants in the same transposon. However, there is no reason for this lack of association. Indeed, like other complex gene arrangements (see below), transposons frequently have a mosaic structure in which highly recombinogenic regions are included (56, 90, 205), so that the acquisition of novel traits is a common occurrence in transposon evolution (256).

Phages. The presence of virulence determinants in phages infecting different bacterial species has been described (Table 2). Also, a phage origin has been proposed for at least some pathogenicity islands (see below). Bacteriophage-associated transduction of antibiotic resistance determinants has been described as well (50, 51, 341, 390). Nevertheless, the presence of antibiotic resistance genes together with virulence determinants in the same phage has not been reported. A possible explanation for this phenomenon is the size requirements for phage DNAs. Phage DNA needs to accommodate inside the head of the phage particle, so that its length must fit a fixed range of sizes. In such circumstances, the gain of some genes must be accompanied by the loss of others. Because of this, the possibility of combining antibiotic resistance genes and virulence determinants in the same bacteriophage is lower than in the case of plasmids, which have less stringent requirements for the incorporation of novel DNA fragments.

View larger version (11K): [in this window] [in a new window]   FIG. 2. Structure of an integron. Integrons are site-specific recombination elements that mediate the acquisition and spread of genes among bacterial populations. Integrons are formed by an integrase gene followed by one primary att recombination site (dark grey box) and several cassettes, each including one gene and one 59-bp recombination site (white square). The transcription of the system is controlled by a strong promoter located upstream of the integron. This structure favors the arrangement of genes in tandem, which are transferred as single elements among bacterial populations.

Another type of gene cassette with a key role in the establishment of the infective process are pathogenicity islands. These DNA regions found in the chromosomes and plasmids of bacterial pathogens are composed of clusters of genes (typically from 15 to 25, although more are possible) which have been acquired by horizontal gene transfer (157). It has been shown that several of these gene arrangements have been introduced into tRNA genes through a phage-mediated transfer mechanism (75). However, a role for insertion sequences has also been suggested in other cases. This is the case for the pathogenicity island required for growth of Y. pestis in iron-deprived environments (162). This island is flanked by direct repeats of IS100, an insertion sequence which is present in the chromosome and plasmid of this bacterial species and thus may contribute to recombination of genes from different replicons.

The presence of sequences with a role in DNA mobility either inside or flanking pathogenicity islands is quite common (157). However, few of these gene clusters have been demonstrated to be mobile. One example is the aforementioned family of pathogenicity islands carrying the gene encoding toxic shock toxin in Staphylococcus aureus (218). Stabilization of pathogenicity islands in the bacterial genome probably requires inactivation of the functions involved in mobility. In fact, the "mobility elements" present in pathogenicity islands frequently carry a large number of mutations, often leading to stop codons (21), which abolish the expression of such functions. Antibiotic resistance determinants have not been detected in pathogenicity islands, but that is an open possibility.

The transfer of gene clusters (contained in either plasmids, transposons, bacteriophages, or gene cassettes) is extremely relevant to bacterial adaptation to novel environments because it allows "bacterial evolution in quantum leaps" (153). As previously stated, both bacterial virulence and antibiotic resistance can be considered strategies to explore and colonize novel environments in which bacterial competitors are scarce. Bacteria will make use of all the tools (mutation, gene transfer, recombination) which allow the high plasticity shown by bacterial genomes (176, 211, 285, 349). In the treated host, adaptation of pathogens will require both antibiotic resistance elements and virulence determinants, so that both adaptive mechanisms must evolve together to produce the explosive spread of virulent and antibiotic-resistant microorganisms that we are now facing.

Environmental signaling in virulence and antibiotic resistance. Bacteria are constantly sensing the environment in order to respond to changes in its composition. In the case of bacterial pathogens, constant cross talk occurs between the microorganisms and their hosts (86) which modulates the expression of several genes in both the bacterial and host cells. Physiological signals such as calcium concentration (372), low pH (128, 296), high temperature (194, 296), and low iron concentration (88, 399) among others, trigger the expression of several virulence genes. However, little is known about whether such signals might also trigger the expression of antibiotic resistance genes.

Nevertheless, we know that the antibiotic susceptibility of bacteria is modulated by several factors which include growth phase (164), pH (89, 203, 226), carbon dioxide (226), temperature (311, 312), bile salts (230), and low iron (305). It is then clear that certain conditions during infection might induce the expression of both antibiotic resistance and virulence genes, and a common regulation of both types of determinants might then occur. Most probably this regulation does not always involve the same regulatory network, but in some cases virulence and antibiotic resistance genes can be part of the same regulon. In this regard, it has recently been described that the MarA protein (from multiple antibiotic resistance) of E. coli regulates the expression of more than 60 chromosomal genes. These genes include not only MDR determinants (156), but also genes with a potential role in virulence, such as those involved in oxidative stress and iron metabolism (33).

In general, a linkage between antibiotic resistance and virulence gene regulation might occur in global regulons, which modulates the expression of stress genes (see below). This type of regulation, coupled with phenotypic resistance of bacteria growing in biofilms, inside cells, and even in resting cells which are resistant to antibiotics, might produce situations of in host resistance at the site of infection that is impossible to predict by routine laboratory susceptibility tests (reviewed in reference 242).

Antibiotics as effectors of bacterial virulence. Common models of bacterial virulence analyze host-pathogen interactions under a defined set of conditions that usually do not include antibiotic therapy. Although the first stages of an infection usually occur without treatment, once a diagnosis is available, most infectious diseases evolve under antibiotic treatment. We want to stress here that antibiotics are not only bacterial killers, but also modulators of bacterial and even host cell transcription of different genes (238, 347). In such a way, antibiotics modulate the interaction between host cells and bacteria (289). It has also been suggested that they may have an important role as modulators of the interactions among natural environmental bacterial populations (100). For instance, quinolones alter DNA supercoiling (5), and the expression of many genes, including those involved in virulence, is dependent on DNA supercoiling (44, 112).

More recently, the role of a natural quinolone on quorum sensing in Pseudomonas aeruginosa has been demonstrated (258, 299). Quinolone treatment may alter the expression of several bacterial genes. In fact, a decrease in the expression of virulence determinants upon exposure to subinhibitory concentrations of quinolones has been described in some cases (62, 361). In contrast, ciprofloxacin and other SOS-inducing antimicrobial agents cause Shiga toxin-encoding bacteriophage induction and enhanced Shiga toxin production from E. coli O157:H7 both in vitro and in animal models (187, 424). Therefore, quinolones clearly affect the expression of virulence determinants, but their role as either enhancers or inhibitors of bacterial virulence depends on the model analyzed and must be established for each pathogenic bacterium. On the other hand, resistance to quinolones as a consequence of mutations in bacterial topoisomerases in some cases alters DNA supercoiling (5, 22, 367), leading to changes in the level of expression of several bacterial proteins, including virulence determinants. For instance, treatment of quinolone-resistant S. aureus with subinhibitory concentrations of quinolones increases the expression of fibronectin-binding proteins, contributing in part to their emergence and maintenance in clinical settings (47).

Other antibiotics besides quinolones can affect the expression of virulence determinants in different ways. For instance, Suerbaum demonstrated that low antibiotic concentrations increased serum sensitivity in E. coli strains protected by the K1 antigen (373), and low macrolide concentrations appeared to reduce the adhesive properties of some gram-negative rods (269). Also, antibiotic treatment may trigger the release of bacterial products, including lipopolysaccharide vesicles (188) and endotoxin (355), that interact with host cells, changing the virulence characteristics of pathogenic bacteria.

Exposure to antibiotics can produce global changes in the metabolism of bacteria, as is possibly the case with Mycobacterium tuberculosis. Exposure to low concentrations of antibiotics induces expression of the alternate sigma factor sigF, which is involved in the regulation of several genes, including those involved in virulence (263). In light of these data, a role of chemotherapy in the persistence of this important pathogen has been suggested. Recently, it has been shown that tetracyclines are iron scavengers (151). Iron deprivation induces the expression of several virulence-associated proteins in bacteria (224, 243), and it is known that free iron is scarce during infection (409). In this way, tetracyclines might trigger the expression of several bacterial genes just as a consequence of their iron-chelating activity.

The influence of subinhibitory concentrations of antibiotics on virulence is then dependent on the bacterial species-antibiotic combination. The response of resistant strains to antibiotics will be different from that of susceptible ones. For instance, if subinhibitory concentrations of antibiotics reduce the expression of virulence determinants, resistance to drugs (even low-level resistance [29]) would restore the original virulence levels. Major efforts must be made to analyze the effect of antibiotics on the metabolism of antibiotic-resistant and -susceptible bacteria. One such approach is the recently published analysis of isoniazid-induced alterations in M. tuberculosis gene expression (7, 415). The authors showed that treatment with this drug produces a dramatic change in the expression of several bacterial genes, some of them related to the drug's mode of action, but others with no apparent relationship to the antibiotic killing mechanism, which may eventually influence the outcome of infection. The current availability of tools for whole-genome expression analysis may allow a better understanding of the role of antibiotics on regulation of bacterial metabolism and thus potentially virulence.

Cross talk between virulence and antibiotic resistance regulons. The environmental regulation of the expression of antibiotic resistance and virulence determinants has already been discussed. The possible cross talk of both types of regulons will now be reviewed. The most relevant example to emerge in the last few years is the effect of quorum-sensing signals on bacterial virulence and the effect of antibiotic resistance on quorum sensing.

Several bacterial species are able to determine the local concentration of bacterial cells, a process known as quorum sensing (37). Quorum-sensing signals trigger the expression of several genes, some of which are involved in bacterial virulence (37, 292, 294). Quite interestingly, it has recently been described that mutations leading to multidrug resistance in P. aeruginosa as a consequence of overproduction of the MexAB-OprM efflux pump also affect the quorum-sensing response (117). This response could be due to a direct effect on the extrusion of quorum signals by this MDR pump in P. aeruginosa (295). More recently, it has also been described that overexpression of the MexEF-OprN multidrug efflux system affects cell-to-cell signaling in Pseudomonas aeruginosa (192).

Since quorum-sensing signals trigger the expression of several virulence-associated genes, such as those involved in production of elastase, rhamolipid, pyocianin, proteases (63), and type III secretion system, antibiotic resistance might strongly influence virulence through the differential processing of quorum-sensing signals in antibiotic-resistant and antibiotic-susceptible bacterial populations. Indeed, recent work in our laboratory has shown that MDR P. aeruginosa mutants are impaired in their virulence properties (334).

Conversely, elements with a role in virulence might also be involved in the regulation of antibiotic resistance. As previously stated, iron deprivation induces expression of the MDR efflux pump MexAB-OprM. Also, the expression of MDR determinants can be induced by salicylate (251). Salicylate is a relevant signal molecule in plant-bacteria interactions (351). It is also an intermediate in the synthesis of different siderophores (104, 346) and is itself a siderophore in Pseudomonas spp. (346). Bacterial siderophores are relevant virulence factors that are produced during infection (243) as a consequence of the small amount of free available iron that is present in the human body (409). Since P. aeruginosa might produce salicylate during infection and this compound induces the expression of MDR determinants, salicylate might thus induce a phenotype of antibiotic resistance for bacteria growing under iron-deprived conditions, such as occur during infection (409).

Mutations in the genes encoding some histidine kinases, which are involved in signal transduction mechanisms in Streptococcus pneumoniae, cause a phenotype of tolerance to the killing ability of glycopeptide antibiotics (281), and these mutants are also less virulent when tested in animal models (384). Deletion of the ciaR gene, which is involved in regulation of competence and cefotaxime resistance in S. pneumoniae, led to a 1,000-fold attenuation of bacterial growth in vivo. Deletion of the vncSR gene pair, which is involved in tolerance to glycopeptide antibiotics in this bacterial species (281), also produces growth attenuation, although the effect is smaller. These signal transduction systems are thus pleiotropic regulators that may influence antibiotic susceptibility, virulence, and even horizontal gene transfer. Therefore, antibiotics may select for variants unable to respond to a number of environmental signals. If that is the case, the environmental signals that induce pathogenicity traits may eventually be altered.

Signal transduction systems also have a role in antibiotic resistance in P. aeruginosa. phoP-phoQ mutants showing increased resistance to aminoglycosides, cationic peptides, and polymyxin have been described (231, 232). PhoP-PhoQ is a two-component regulatory system that has been studied mainly in S. enterica serovar Typhimurium. In this bacterial species, PhoP-PhoQ regulates the expression of at least 40 genes in response to magnesium concentrations (360) and has an important role in both virulence (139, 266) and resistance to polymyxin (155). Although a role of P. aeruginosa PhoP-PhoQ in virulence has not yet been demonstrated, it is highly expressed under magnesium starvation (232), which indicates a role similar to that previously described in the case of S. enterica serovar Typhimurium. PhoP-PhoQ may then coordinately regulate both virulence and antibiotic resistance in response to the magnesium concentration.

We have addressed some examples of cross talk between antibiotic resistance and virulence regulons. However, genes encoding both types of determinants might also be part of the same regulon. As previously stated, this is the case for the Mar regulon in E. coli and S. enterica serovar Typhimurium, which is involved in multiple antibiotic resistance (156), superoxide resistance (106, 177), organic solvent tolerance (18), and bile salts resistance (230), among others. The same situation probably occurs for several other MDR determinants, so that the possibility that the physiological signals that bacteria receive during infection might trigger a global response including the regulation of both antibiotic resistance and virulence genes should be carefully analyzed.

THE CROSSROADS OF VIRULENCE, EPIDEMICITY, AND ANTIBIOTIC RESISTANCE

Top Previous Next References

 
The pathogenic relevance of a bacterium relies mainly on two different properties, virulence and epidemicity (the ability to produce epidemics). A compromise between the two properties is required to cause a major disease because a nonvirulent bacterium obviously will not produce any disease, whereas a nontransmissible bacterium might cause a dangerous disease but only in a limited number of people. Since bacteria are under antibiotic pressure during infection (see above), highly efficient pathogens must be not only virulent and epidemic but antibiotic resistant as well. The problem from the microorganisms' perspective is whether or not the acquisition of one of these traits could have a detrimental effect on the expression of the others. In such a case, an "optimization strategy" will be required. We will review whether the acquisition of these abilities is detrimental to the expression of the others and, conversely, reinforces the possibility of acquisition of a full proinfective repertoire, virulence, epidemicity, and antibiotic resistance.

Although it is difficult to find examples of the effect of antibiotic resistance on the virulence of clinical bacterial isolates, epidemiological data support the idea that, in some cases, antibiotic-resistant organisms may show a decrease in pathogenicity. For instance, in urinary tract infections, fluoroquinolone-resistant E. coli are easily found in cystitis but are isolated at very low frequencies as causative agents of pyelonephritis (54). Recent studies also suggest that penicillin-resistant S. pneumoniae strains may be less pathogenic than susceptible ones, at least in animal models (20, 234), and isolates from cases of bacteremia are frequently less resistant than those isolated from mucosal infections and even in carriers (150, 175). In this case, a possible explanation is that the capsulation involved in serum resistance eventually interferes with acquisition of foreign DNA encoding resistant penicillin-binding proteins.

Nevertheless, the pathogenic process might provide an advantage for the bacteria, such as permitting access to exclusive habitats. For instance, a reduction in the rate of protein synthesis because of a ribosomal resistance mutation may be deleterious for a bacterial organism competing with the wild-type susceptible strain in a rich medium, where the effect of such a deficiency is maximized, while under slow-growth conditions, the difference would be minimal. The expression of genes involved in pathogenicity is frequently regulated by stationary-phase signals (160, 207), so that the slow growth due to a reduction in protein synthesis might eventually trigger the expression of genes involved in pathogenesis. In this case, the growth defect produced by antibiotic resistance could be compensated for by the increased virulence of the pathogen, and bacteria could easily accumulate compensatory mutations which restore their original pathogenic abilities totally or in part (213).

Although antibiotic resistance usually reduces bacterial fitness, the opposite situation might also occur. For instance, the bleomycin resistance gene contained in transposon Tn5 confers improved survival and growth advantage on Escherichia coli (55). The increase in the virulence properties of antibiotic-resistant bacteria is a disturbing possibility, because in these cases, revertants that become susceptible by means of mutations and loss of antibiotic resistance plasmids may be at a selective disadvantage relative to the ancestor resistant population, so that the resistant population will prevail even in the absence of antibiotic selective pressure.

In general, antibiotic resistance obviously increases the overall pathogenic potential of the organism in the treated patient, because the susceptible organisms are killed by the antibiotics and the infection does not progress. For instance, the mortality rate of S. pneumoniae meningitis in South African children is significantly higher for antibiotic-resistant strains (133, 134). However, in the absence of antibiotics, antibiotic resistance may influence pathogenicity in different ways, depending on the antibiotic resistance determinants involved and the possible accumulation of compensatory mutations.

Several reasons can be given in favor of the existence of an evolutionary link between antibiotic resistance and host-to-host transmission. The diminution in bacterial diversity due to antibiotic treatment will result in overgrowth of resistant bacterial populations that are in the minority under normal competitive circumstances. Antibiotic treatment is then a bottleneck that is crossed by the few antibiotic-resistant bacteria that are present in the population. However, once this bottleneck is crossed, the best colonizers among the remaining antibiotic-resistant bacteria will have an advantage for recolonizing the host. If the challenge is repeated several times, it could be expected that a certain within-host evolution of bacteria to optimize colonization will occur (412). Considering that all members of the host population are very similar in their ability to be colonized by each particular organism, success in colonizing the host will be reflected in a corresponding success in between-host transmission ability. Large numbers of resistant cells within the host should facilitate host-to-host transfer, particularly to other treated hosts. This perspective may have some exceptions. Some bacterial populations are strictly linked to particular environments, including a dependence on other local bacterial populations (41). That situation is not infrequent in mucosa-associated communities. In these cases, the antibiotic-resistant population may remain confined and even disappear under antibiotic therapy just because the other members of the community are eradicated.

The possibility that a resistant population will be enriched under antibiotic treatment thus requires the ability of this species to independently exploit the available habitat. This ability is common among organisms capable of crossing ecological barriers, including the more epidemic ones. In serial-passage experiments with mixed populations, the strain with the greatest number of cells in the transferred inoculum has a selective advantage. The driving force may be within-host competition and selection for increased parasite growth rate (114). In this regard, the ecological abundance of specific clones in natural populations of Staphylococcus aureus might be linked to the virulence of such clones (99). The clones should then have been selected because they are good colonizers as well as being highly transmissible and highly virulent (99, 221). This statement is true in general, except perhaps for organisms with high virulence, such as Mycobacterium tuberculosis, for which a small inoculum may be sufficient to cause the illness.

Epidemicity ensures constant high multiplication rates that may be needed for the acquisition of resistance. Therefore, these bacteria are simultaneously more prone to becoming resistant, to be selected for by antibiotics, to increase their absolute numbers, and consequently to be transmitted efficiently. There is an obvious corollary: antibiotics should be prescribed (57) in a way that ensures eradication of the bacterial pathogen (as said by Paul Ehrlich, "hit hard and hit early"), as any survival gives the organism an opportunity to evolve and spread in a more efficient way.

The relationship between epidemicity and pathogenicity may also help to modulate the development of resistance. Within-host growth usually induces the expression of virulence determinants in several pathogens, thereby increasing their virulence. Increased between-host bacterial transfer may also increase virulence. It is also well known that serial passage of a given organism among susceptible hosts frequently increases its pathogenic power in the new host, although it is attenuated fo